Fraunhofer-Gesellschaft

Publica

Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

The accurate and early effort estimation of web applications

 
: Ruhe, M.
: Rombach, H.D.; Wieczorek, I.; Jeffery, R.

Kaiserslautern, 2002, XI, 174 pp. : Ill., Lit.
Kaiserslautern, Univ., Dipl.-Arb., 2002
English
Thesis
Fraunhofer IESE ()
cost estimation; size measurement; web application; case study

Abstract
This thesis investigates whether the application of the CoBRATM method is appropriate for cost estimation of web applications. Web applications became of interest to researchers in the last couple of years. The success of web applications used to be determined by being first-to-market. This has changed towards quality web applications. Quality software applications require rigorous processes. Cost estimation is one of the crucial steps at the very beginning of a new software project. Accurate cost estimates are an essential element for being able to provide competitive bids and remaining successful. Only a few researchers have addressed cost estimation of web applications so far. A lack of detailed research results in this area has been identified.
The CoBRA method has only once been applied in a large traditional software house mainly developing MIS applications in Germany [Briand et al., 1998]. The current work applies the CoBRA method in the context of Allette Systems, a small Australian company that focuses on web development. The CoBRA method has been refined and tailored for the application in a small organization and a new application domain. This also includes the investigation of an appropriate size measure for web applications.
This thesis presents a detailed description of the CoBRA method. The transparency of the method increased by documenting all steps performed when developing the Web-CoBRA model. Weaknesses of the method are identified and addressed within this thesis. A refinement of the CoBRA method has been developed.
The evaluation of the developed Web-CoBRA model includes data on twelve past projects and their analysis in terms of predicition accuracy using cross validation techniques. The estimates derived from the Web-CoBRA model are significantly better than the ones derived from Allette Systems former informal estimation method. Their previously obtained mean magnitude of relative error of 0.37 is decreased to 0.17 using the Web-CoBRA model.

: http://publica.fraunhofer.de/documents/N-14657.html